Principles of Neuro Neur30003 Notes PDF

8.3 Learning & Memory – mechanisms Hebbian modification Two forms of Hebbian modification: o Long term potentiation (LTP): Long term activity dependent plasticity that satisfies the criterion that synapses strengthen when pre- and post-synaptic neurons are active at the same time i.e....

8.3 Learning & Memory – mechanisms Hebbian modification Two forms of Hebbian modification: o Long term potentiation (LTP): Long term activity dependent plasticity that satisfies the criterion that synapses strengthen when pre- and post-synaptic neurons are active at the same time i.e. when stimulated CA1 neurons in tetanus, they got a much bigger response with a single stimulus afterwards. o Long term depression (LTD): synapses weaken when pre- and post-synaptic activity is not simultaneous LTP & LTD in hippocampal neuron Tetanus: high frequency stimulation - gets its name from the Clostridium tetanus, where all inhibitory synapses (esp those with glycine) are blocked because tetanus toxin blocks release of neurotransmitter. - Any reflex activity or activation of sensory stimulus into spinal cord causes massive overreaction of motor neurons, causing tetanus and muscle contracts maximally. LTP: - Neurons that were tetanically stimulated had much larger response. This effect lasts >1 month. LTD: - Results in much smaller response because stimulating at a low freq, not activating the post-synaptic neuron to fire AP - No change to neuron that doesn't receive the direct input. These 2 phenomena are enough to produce a memory - If they were there by themselves, and no other mechanisms operating, associations would always drive memories towards 1 single memory - synapses strengthened keep strengthening and weakened synapses keep weakening, pushing network to one pattern of activity. - Sleep: allows this to be reset, so relative rather than absolute changes in synapse strength is imposed A post-synaptic mechanism for LTP  Glutamate excites AMPA receptors and unblocks NMDA receptors for next EPSP  Glutamate receptors become more sensitive to glutamate – releasing a certain amount of glutamate elicits a greater response  AMPA receptor is 1st channel opened (let Na+ in, K+ out)  At rest, NMDA isn't opened because it has Mg2+ sitting in pore. When depolarised, Mg2+ comes out  NMDA has slower action kinetics, and slower closing kinetics (opened for longer and Ca2+ influx leads to larger and longer depolarisation).  Ca2+ entering via NMDA receptors activates Ca2+- dependent kinases (e.g. CaMKII)  Kinases phosphorylate AMPA receptors (makes them more sensitive to glutamate) and cause insertion of AMPA receptors into postsynaptic membrane from intracellular receptor pool  Later, glutamate releases are more effective.  Ca2+ can enter via NMDA & voltage-gated Ca2+-channels too  So, LTP is occurring because activation of kinases makes AMPA more sensitive & puts more AMPA receptors into membrane. NMDA is needed for Ca2+ influx, but not actually what's responding to create LTP.  Ketamine blocks memory formation because it blocks NMDA receptors. Memory in a molecule: Calcium-calmodulin dependent protein kinase II (CaMKII) - In spine cytoplasm: associated with postsynaptic density - Associated in rings of 10 subunits/associated enzymes, each of which can be activated through calcium- calmodulin opening - Ca2+-calmodulin disinhibits kinase activity - CaMKII autophosphorylates - Phosphorylated CaMKII constitutively active until dephosphorylated by a phosphatase - Memory stored in number of CaMKII molecules within ring that are phosphorylated - So, intracellular mechanism that puts AMPA receptors into the membrane also makes its own molecular memory inside synaptic spine involved.  - Individual spines are changed, but not all of them. - Once phosphorylated, hinge is open. - phosphatases that can take phosphate group off and cause hinge to close - causes LTD. - This is also ca2+ dependent, respond to smaller increases of conc of ca2+ than the amount of ca2+ needed to activate CaMKII. Regulation of protein synthesis  Phosphorylation, receptor sensitization, increased receptor number – all time limited  For long term memory need protein synthesis  One mechanism identified in invertebrates involves phosphorylation of CREB (cyclic AMP Response Element Binding protein), which regulates gene expression NMDA receptors are not the only source of Ca2+ for plasticity  Back propagating dendritic APs can open voltage-dependent Ca2+ channels in dendritic spines  Back propagating dendritic AP: Intiating an AP at AIS, the AP doesn't only go down axon, but also back into soma. If dendrites of neuron have voltage-dependent Na+ and Ca2+ channels, it'll open these channels  Ca2+ entry via NMDA plus voltage-dependent channels greater than sum of separate routes of entry by themselves  Can also ↑cytoplasmic Ca2+ via release from intracellular stores – e.g. metabotropic glutamate receptor mGluR1 causes Ca2+ release from endoplasmic reticulum  Timing of Ca2+ entry is the key So we can get 3 diff ways glutamate can activate ca2+ entry: Release from stores in mGluR, NMDA receptors, Back propagating dendritic AP Back propagation of an AP in a dendrite  Back-propagating dendritic APs activate voltage-dependent Ca2+ channels  Close association of back propagating APs and EPSPs lead to either LTP or LTD  Depends on timing Timing and LTP/LTD  Timing of the back propagating AP and the EPSP is critical  If AP in post-synaptic neuron precedes EPSP  LTD current has not been involved in the generation of that action potential  If AP follows EPSP  LTP suggests that the EPSP has contributed to the generation of the action potential  Recording from 2 neurons at the same time Spike timing plasticity  When AP in post-synaptic neuron occurs before an EPSP total increase in intracellular Ca2+ is small  Because increase in Ca2+ conductance due to AP is very short, so doesn't add onto the ca2+ signal you get from NMDA channels opened during synaptic potential  Activates Ca2+-dependent phosphatase  LTD  When NMDA-receptor mediated EPSP occurs before post-synaptic AP increase in intracellular Ca2+ is much larger  Because NMDA receptors remain open for quite some time after glutamate  AP that's back propagating coincides with open time of NMDA channel  So ca2+ is coming in through NMDA channel, and voltage-dependent ca2+ channels  Ca2+ signal is much larger, which activates Ca2+-calmodulin, which activates CaM Kinase II  LTP  9.1 Time Place Space Hermann Ebbinghaus: A psychophysicist, used a process of getting people to learn novel info (e.g. nonsense syllables) and testing you after periods of time. Learning and retaining information is, typically, limited by how much information is involved, how long you need to retain it, and how often you rehearse the information. Over 100 years ago Herman Ebbinghaus quantified this well- known phenomenon as an exponential decay function. (e.g. 20 mins after learning, we forget about 1/2 of info) Ebbinghaus - Wanted to find out where in the brain a memory is, & what's changed - Engram: the change in brain that now represents some new info that you can access - Trained rats to do things (run through mazes, etc.) and then cut out bits of brain to see which part retained memory - Removed a chunk of cerebral cortex, and had to remove 1/2 of brain before he got performance deficit (but still had some residual performance) - Even removing whole cerebral cortex: some residual performance - Morris water maze: rodents can swim in opaque liquid, can find a platform, and in the future, can find it quickly (spatial memory) 2 ways animals appear to learn about their spatial environment and their location and orientation within it: 1. navigate by dead reckoning - by keeping track of the direction, duration and distance of travel in straight lines and summing these vectors to calculate current position in relation to the starting position 2. detailed exploration of environments allows spatial information to be integrated into cognitive maps, by accruing lots of info from dead reckoning (diff routes form a model). Cortical lesions studies in rodents by Karl Lashley in the 1950s: tried to find the location of memory traces,  revealed there wasn't a discrete cortical location for memory; very large lesions were needed to eliminate memories of tasks learned by rats (such as maze solutions). Why were Lashley’s cortical lesions so ineffective at diminishing learned performance in tasks like conditioned reflexes and maze running? o Strong evidence for specific functional location of a type of memory in a brain sub-region cam for human neuropsychology o H.M. showed that the hippocampus is essential for the consolidation of explicit episodic memory: - Had surgery to remove two hippocampi (medial temporal structures). - lost his capacity to learn new information for long time - anterograde amnesia. - Could remember things prior to surgery - In one lobe, it wasn't a complete lesion, so no complete memory loss. o Loni Sue Johnson - Suffered a viral infection and nearly died, recovered but caused neurodegeneration of both hippocampi - Caused anterograde amnesia (no new memories) and also retrograde amnesia - Hippocampus is important to form memories, and retrieve memories previously formed.  In rodents, spatial navigation and spatial memory was demonstrated (e.g. by lesion studies) to be hippocampal dependent.  In humans however, it seemed that he hippocampus was mainly concerned with consolidation of episodic memory (patient H.M. lived in a small window - about 2 minutes of the present, his only longer term memory was what he wrote into notebooks)  Because it is a good slice preparation (the connectivity between neurons is preserved after cutting), the rodent hippocampus has been used to study synaptic transmission. It was in hippocampal slices that neuroscientists Bliss and Lomo discovered LTP. It makes sense that hippocampal synapses are capable of profound and long-term changes in strength, as this would seem to be a good candidate for how memories are formed. Primate hippocampal anatomy  LGN is close to where hippocampi are.  Hippocampus is a cortical structure, and at very edge of cortex.  Most of cortex has 6 layers, and at edges, becomes less layers. Hippocampus has 3 cellular layers. Rodent hippocampal connections, esp. CA3 to CA1, have been a favorite preparation to study synaptic plasticity. The “canonical” hippocampal circuit receives input from, and projects back to, the Entorhinal cortex. o Perforant path: input to hippocampus from entorhinal cortex o Which makes excitatory projections onto granule cells in the dentate gyrus. o These granule cells send excitatory projection to CA3 (cornu ammonis) region o Which send excitatory projection to CA1 neurons o They project back out again. The NMDA receptor channel can open only during depolarization - Normal excitation just excites AMPA receptors, allows Na+ entrance. - If post-synaptic cell is depolarised at the time, also get NMDA receptor excited because it loses its Mg2+ block. Hippocampal neurons are strongly modifiable by LTP. In the 1970’s it was discovered, by recording from hippocampal neurons in free behaving rats, that some neurons encode a small region of the rat’s environmental position (or “place”). It now seems that many if not most hippocampal neurons are place sensitive. These cells only fired when in a particular place in environment - so place cells that responds to where cell was were accidently discovered 1st thought to be memory of location, later seen as present location encoding, and part of a system of space-encoding neurons:  Place cells  Grid cells,  Boarder cells: active when you reach a border  Head direction cells  Speed cells Add synaptic plasticity for fast declarative memory Place fields/cells - Accidental discovery by John Keefe when unit recording was made measuring hippocampal neurotransmission in awake freely behaving rodents - Are neurons that increased their firing rates when the animal was in a specific regions of it’s environment. - An array of place fields, some D-V hippocampal order but not topographic – ie neighbouring cells were not necessarily nearby place fields – but combination of place cells provided a pattern unique for each location. - This was looking like a cognitive map as proposed by the psychologist Tolmen in 1948 - But H.M. had strongly suggested a role for hippocampus as essential for declarative episodic memory. Clues: Hippocampal neurons that respond as place cells also response to sensory stimuli - touch, taste, and time. So a place field is also what happens in a place.  Unlike semantic memory, episodic memory is tied to space.  Even early work on place cells had shown they not only encode where the rat is, but where it had been recently.  Sequences of space cell activations were replayed at rest and during sleep, suggesting the map can be used to imagine / plan journeys. Is the map set or does it reform with the exploration of new environments?  Initial experiments with a small sample of place cells suggested a fixed hardwired (predetermined) space map, but more complete studies showed that many place cell fields take several minutes to settle, and this depended on the rats attention to, and exploration of, the new environment.  Experience reformates the spatial map. Because some neurons re-map so rapidly, it is suggested that there may be a set of skeletal templates that can be selected then further refined as the specific of new environment are encountered.  Place cell was used again to 'remap'/activated in diff places What would happen if you blocked NMDA-mediated LTP? o Synaptic transmission still happens because AMPA receptors are unaffected, but no LTP o No reassignment of place cells: stopped them adapting to new environments  o Assignment of place cells in diff environments is due to synaptic plasticity. Place cell activity is coordinated during and after exploration o The activities of place cells, during exploration and rehearsal, occur in the context of intrinsic rhythmic, coherent hippocampal activity - theta rhythm. o has a number of postulated functions is but its significance is not currently understood o It has been known for some time that hippocampal neurons fire in accord with a 6 - 10Hz (theta) rhythm, which is dominant during exploratory phase of navigation. o This rhythm seems to be important in consolidating spatial environ. Esp in sleep, this rhythm gets stronger o Consolidation is dependent on the sharp wave activity that spreads as ripples through the Hippocampus and its entorhinal connections at rest and sleep. The sharp-wave activity is optimal for LTP and spreads via entorhinal cortex to cerebral cortex – this process has a long history of theoretical and experimental support for being the basis by which learning is transferred from Hippocampus, becoming consolidated as cortical representations. o May relate to conscious recall of spatial information. The sharp-wave ripples also seem to be used in imaginary journeys or rehearsal - once a map is established we have access to mental travel time estimation Where do the signals that contribute to the hippocampal place fields come from?  Despite predictions, place fields were not generated from intra-hippocampal connections – isolating CA1 from CA3 and DG didn't have much effect.  The entorhinal cortex showed strong modulation by place with clearly silent regions separating place fields. Unlike place cells (that fire when the rat is in a particular location) grid cells fire when a rat is in a multiple of a particular distance.  May-Britt and Edvard Moser showed that the place cells get input, and get reassigned their place fields, from a collection of grid cells in the entorhinal cortex  These cells formed a grid (triangular or hexagonal) they are non-topographic but denser space dorsally and wider ventrally the mesh size varies from about 30cm to over 3 meters (so wide earlier closely spaced neuronal recordings missed the grid).  The sequential firing of cells representing larger and larger grids give the rat a reference system for its location in its environment.  When we go from environ to environ, we probably activate a new grid. Probably associates spatial features and visual features we know about environ too  Importantly, place cells also receive input from sensory systems, and thus encode not only location but also events and objects associated with that location - a place field is also what happens in a place and is thus also a type of episodic memory.  Place cells not only encode where the rat is, but where it had been recently; the sequential pattern of place cell activation was replayed when animals slept.  suggested that this is a process of consolidating the spatial memory  Once consolidated (and probably stored outside the hippocampus) a spatial memory can be recalled when returning to the environment, or even perhaps recalled into the imagination to plan journeys. Head direction cells  Tells hippocampal map which way you’re oriented  A 2nd type of cell defining the animals place in the environ appears in regions downstream of hippocampus (entorhinal cortex, subiculum)  Cell fires when animal is facing a particular direction in an environment (relative to environ, not compass!) How do grid cells encode a specific combination of place cell identities for each new environment? Unlike place cells, there's a dorso-ventral gradient of grids (somewhat topographic). If you combine diff sets of grid cells, can uniquely specific thousands of diff environs at diff scales. E.g. there may be a combination of grid cells for lecture theatre, then a diff combination for tram stop Navigation requires a combination of a) “dead reckoning” or self-referenced movement from a known location, and b) the generation of landmark-based cognitive maps. Maybe that's how other types of memory form in general - e.g. semantic knowledge may be “mapped” into conceptual and relational categories It is thought that the formation of a detailed maps relies on repeated experiences with self-referencing explorations, the same way as semantic memories may become context-independent with repetition of episodic memories concerning a semantic relationship